50 THE MEDUSAE. 



1. Cunanthidae ; gastric pockets undivided, equal in number to the ten- 

 tacles ; with or without peripheral canal system ; with or without otoporpae. 



2. Aeginidae ; gastric pockets adradial (in adult), usually twice (or four 

 times) as numerous as the tentacles (an exception to the rule is seen in Aegina 

 aliernans) ; with or without canals ; without otoporpae so far as known. 



3. Solmaridae; without gastric pockets; with or without canal system and 

 otoporpae. 



Besides the light which the collection throws on the classification of the 

 group, it is very important, since it contains examples of two methods of re- 

 production : — internal budding, and development of a secondary, parasitic 

 generation of medusae. Since the first of these has never been conclusively 

 demonstrated, the discovery that a form of internal budding does occur in 

 this order is of unusual interest, even though only the earlier stages in the 

 process can be traced. The development of a parasitic generation directly 

 from the egg in the body of the parent host has already been described by 

 Metschnikoff ('SG''), under the name "sporogony," and by Stchelkanowzew 

 (: 06), for Cimhia proboscidea Metsch. The process is so rare, and so remark- 

 able on account of the occurrence in the one species of two alternating gener- 

 ations of medusae of different anatomical characters, that a repetition of 

 the observations for another family (Solmaridae) is of great interest. I 

 should add that in addition to the modes of development outlined above there 

 have been described for various Narcomedusae direct development from 

 the egg, budding from stolons parasitic in various medusae (Cunina, Cunoc- 

 tantha), and the ordinary external formation of medusa buds (Cunoctantha, 



Browne, : 06). 



Cunanthidae Haeckel, 1879. 



sens. em. Maas (: 04% : 04'). 



Narcomedusae with undivided radial gastric pockets, equal in number to 

 the tentacles. 



Of the five genera distinguished by Haeckel under this family, Maas has 

 shown that two only, Cunoctantha and Cunina, really belong here. Of the 

 three remaining, Cunarcha and Cunissa belong to the Aeginidae, since they 

 have divided gastric pockets twice as numerous as the tentacles, while the 

 third, Cunantha, seems almost certainly to be based on larval stages of either 

 Cunoctantha or Cunina. According to Maas's classification, which I have 

 adopted, Solmissus, classed by Haeckel ('79) under the Solmaridae because it 

 has neither canal system nor otoporpae, must be referred to the present family 

 because of the conformation of its gastric pockets. 



